Time scale of the excitation of electrons at the breakdown of the quantum Hall effect

Transcription

1 Time sale of the exitation of eletrons at the breakdown of the quantum Hall effet B. E. Saǧol, 1 G. Nahtwei, 1 K. von Klitzing, 2 G. Hein, 3 and K. Eberl 2 1 Tehnishe Universität Braunshweig, Mendelssohnstrasse 2, D Braunshweig, Germany 2 Max-Plank-Institut für Festkörperforshung, Heisenbergstrasse 1, D Stuttgart, Germany 3 Physikalish-Tehnishe Bundesanstalt, Bundesallee 100, D Braunshweig, Germany Reeived 12 Marh 2002; published 2 August 2002 We have investigated the time sale of the exitation of eletrons leading to a transition from the quantum Hall QH state to the dissipative state in the two-dimensional eletron systems of GaAs/AlGaAs heterostrutures. The measurements were performed by applying nanoseond eletri pulses to the urrent ontats of QH devies having various eletron mobilities ( H m 2 /V s) at integer filling fators 2, 4, and 6. The breakdown of the quantum Hall effet QHE ours when the applied pulses exeed a ritial pulse width t p that is a funtion of pulse amplitude, magneti field, and eletron mobility (0.4 t p 18 ns). By applying a simple drift model, we obtain an estimate for the ritial drift lengths, whih vary from about 1 to 25 m. No general differenes of the response to short eletri pulses have been found between Hall bar and Corbino devies. We interpret the ritial times as drifting times between inelasti sattering events, and the ritial lengths as related to the inelasti sattering lengths in of eletrons ausing the QHE breakdown. Charateristi dependenes of the ritial drift times and lengths on the amplitude, filling fator, and the mobility have been observed and an be attributed to impurity-assisted inter-landau-level tunneling. DOI: /PhysRevB PACS number s : Jn, J I. INTRODUCTION In the integer quantum Hall effet QHE the longitudinal resistane xx vanishes, while the Hall resistane xy has plateaus with values xy h/ie 2, where i is an integer number. 1 If the urrent density of the two-dimensional eletron system 2DES exeeds a ritial value, the QHE breaks down 2,3 exitation of eletrons, and if the urrent density is redued to subritial values, the QHE is reovered relaxation of eletrons. Experiments with multiterminal quantum Hall devies 4 7 have shown that both exitation and relaxation of eletrons develop over a ertain drifting distane of eletrons. For a review on the QHE breakdown, see Ref. 8. The experiments probing the relaxation of hot eletrons bak to the QH state 6 yielded a linear relation between the mean free path mfp and the average drift length D between two inelasti sattering events. Dividing these drift lengths by the drift veloity, it was possible to estimate the relaxation times rel in the nanoseond range, whih are just the average drifting times of hot eletrons for the relaxation. As mfp at temperatures below 4 K is dominated by elasti sattering at impurity potentials, the relaxation of exited eletrons bak to the QHE was attributed to inter-landau-level transitions, supported by Coulomb sattering at impurities. This explanation of the QHE breakdown was based on the so-alled quasielasti inter-landau-level sattering model, whih was proposed by Eaves and Sheard. 9 The reverse experiments of the exitation of eletrons from the QH state to the dissipative state also showed a lear orrelation between the mobility or the sattering density and the spatial evolution of the QHE breakdown along a onstrition in the urrent path. 7,10 With inreasing density of satters dereasing mobility, the avalanhelike breakdown beame steeper along the drift diretion. However, it was not possible to estimate the orresponding exitation times ex in these experiments. By applying a phenomenologial model for the transitions between the two Landau levels above and below the Fermi level, the avalanhe ould be simulated. 11 This model is based on the well-established assumption of the presene of potential flutuations mainly due to harged impurities. 4 These flutuations provide loalization enters for eletrons and holes in the Landau levels and lead to loally enhaned eletri fields. On the basis of these assumptions, the breakdown of the QHE was explained by the field-dependent evolution of ompressible islands, whih merge with marosopi areas onneting the sample edges at the breakdown of the QHE. 12 Involving the evolution of these ompressible islands in time by inter-landau-level transitions, as done in Ref. 11, one an obtain the length sale of the breakdown along the drift diretion of the eletrons. Exitation times ex average drifting times of the eletrons between inelasti sattering events of the order of the relaxation times rel ( ex rel ) ould be dedued. However, these values of ex and rel have been estimated by assuming a homogeneous urrent flow over the width of the sample, an assumption that is still a subjet of disussion see, e.g., Ref. 8. In order to measure the time sale of exitation ex diretly, we applied retangular pulses to the 2DES of a Hall bar with pulse widths t p (0.3 ns t p 0.4 ms). We found that the breakdown ours only after exeeding a ertain ritial pulse width of the order of nanoseonds, whih is a funtion of magneti field and pulse amplitude. 13 In this work we present the experiments, whih are performed by applying pulses with pulse widths in this ritial nanoseond range to the urrent ontats of QH devies, having different eletron mobilities and geometries. After introduing the experimental methods in Se. II, we present the experimental results and disuss the similarities and differenes between Hall bar and Corbino geometries. In Se. IV a simple drift model inluding harged satterers and potential flutuations is introdued and disussed with regard to the /2002/66 7 / /$ The Amerian Physial Soiety

2 SAǦOL, NACHTWEI, VON KLITZING, HEIN, AND EBERL TABLE I. Important devie parameters. Eletron density n s and Hall mobility H were determined from transport measurements at 1.5 K. Devie Geometry Ch. width m n s m 2 experimental results. Finally Se. V inludes the summary of our work. II. EXPERIMENTAL METHODS H m 2 /V s H1 Hall bar H2 Hall bar H3 Hall bar C1 Corbino C2 Corbino FIG. 1. Shemati view of the experimental setup to measure the pulse-indued breakdown in Hall bar a and Corbino devies b. Retangular pulses with onstant duty yle are applied between the soure ontat S and the ground. At the Hall bars the average longitudinal resistane xx, at the Corbino rings the average longitudinal ondutivity xx were measured. The Hall bar devies H1, H2, and H3 were fabriated from three different GaAs/GaAlAs wafers with a hannel width of w 272 m and distane l 272 m between potential ontats. The Corbino devies C1 and C2 were strutured on the same wafer, with the same inner radius r m, but having different hannel widths of w 100 and 200 m r and 300 m. The important devie parameters are shown in Table I. The experimental setup is realized as shown in Figs. 1 a and 1 b : a oaxial able with a 50- impedane-mathing resistor was onneted in the immediate viinity of the samples. Retangular pulses V A (t), with amplitude A, were applied between the soure ontat S and the ground. The pulse width t p and the period T were inreased simultaneously, keeping the duty yle equal to t p /T onstant and therefore keeping the average applied voltage onstant for eah measurement urve. Another resistor was attahed in series with the devies between the drain ontat D and the ground in order to measure the average urrent I s passing through the devie. The averaging of the time-dependent voltages was performed by using a nanovoltmeter with an input time onstant in set to in t p, T. The setup was arefully alibrated by measurements at ohmi referene resistors with known resistanes. All of the experiments were done at 1.5 K, and magneti fields B up to 8 T were applied perpendiular to the 2DES. At the Hall bar devies the average longitudinal voltage V x and the average urrent passing through the sample I s were measured. The average longitudinal resistivity was alulated aording to xx V x I s w l, where w is the width of the sample, and l is the distane between the suessive measurement probes. At the Corbino devies, the average urrent passing through the samples I s was measured, and the average longitudinal ondutane was alulated as xx I s ln r 2 /r 1, 2 V SD 2 where V SD is the average applied voltage between the soure and drain ontats, and r 2, r 1 are the outer and inner radii of the Corbino ring. III. EXPERIMENTAL RESULTS A. d measurements In order to haraterize our samples, we measured the soure-drain urrent I s at Corbino and Hall bar devies and the longitudinal voltage V x at Hall bars only as a funtion of B, for fixed d values of the soure-drain voltage V SD.By using these data and the formulas 1 and 2 we alulated the SdH traes for xx of Hall bars and xx of Corbino devies. From these traes we determined the eletron onentrations n s and the Hall mobilities H of the QH devies. Figure 2 shows two examplary traes for xx and xx of a Hall bar and a Corbino devie, respetively, whih were fabriated from the same wafer. Further, it was neessary to determine the breakdown onditions at the integer filling fators, where the ritial urrent of the breakdown reahes its maximum value. Several urrent-voltage (I-V) traes at different magneti fields were obtained in order to determine the integer filling fator positions and the orresponding ritial d voltages V SD. Figure 3 a shows suh an I-V urve of a Corbino ring (C2) at filling fator 4. The ritial d voltage is V SD 0.57 V for the up sweep and V SD 0.53 V for the down sweep. The ourrene of suh a hysteresis in the I-V urves of the QHE breakdown 3,14 is rather typial and usually attributed to thermal instabilities of the eletroni system near the ritial point. B. Pulse measurements Figure 3 b shows the pulse-indued breakdown urves of this Corbino devie (C2) for different pulse amplitudes. In ase the applied pulses have amplitudes lower than these ritial voltages, e.g., for the amplitude A 0.5 V, no breakdown is observed. But as the amplitude is inreased from

3 TIME SCALE OF THE EXCITATION OF ELECTRONS... FIG. 2. Two examplary traes for xx of a Hall bar (H3) and xx of a Corbino ring (C2), whih were fabriated from the same wafer. to 0.6 V, above the ritial voltage, the QHE breaks down at a ertain ritial pulse width, at whih the average longitudinal ondutivity xx starts deviating from its initial xx 0 value. If the amplitude is inreased to even higher values, the breakdown ours more abruptly and at smaller ritial pulse widths. The duty yle in these measurements is 20% so that the average applied voltage V A A/5 is muh smaller than V SD. Therefore the amplitude is the ritial parameter for the breakdown but not the average voltage. Similarly, Fig. 3 shows the pulse-indued breakdown measurements on H3 at 4 for different pulse amplitudes. Eah urve is measured by applying pulses with a onstant amplitude A, and a onstant duty yle of 20%. Both the average resistivity and the average urrent are plotted as a funtion of the pulse width. For this filling fator, the ritial d breakdown voltage is V SD 0.73 V. Therefore, as expeted, there is no breakdown for an amplitude of A 0.6 V, suh that the average resistivity is xx 0, and the average urrent stays onstant for all pulse width values. However, as the amplitude is inreased to A 0.9 V, above the ritial d voltage, the QHE breaks down at a ertain ritial pulse width. This an be observed on both sets of the graphs in Fig. 3, with the inrease of xx, and orrespondingly with the derease of the average urrent. As the amplitude is inreased further, the breakdown starts at smaller pulse widths, i.e., the higher the amplitude, the lower is the ritial pulse width t p. In other words, even if the amplitude exeeds V SD, there is a ertain pulse width range, in whih no breakdown takes plae. As stated before, these experiments were performed with a onstant duty yle, suh that both the pulse width and the pulse period were hanged simultaneously. Thus, the ourrene of dissipation for pulse widths t p t p is not a onsequene of the hange of the average soure-drain urrent as I s onst holds, but of the proess during the pulse duration itself. To prove this, we have studied the behavior of the samples for different duty yles i.e., different pulse-topause ratios. Figure 4 shows typial measurements on the Hall bar and Corbino geometries at filling fator 4 for different duty yles, i.e., for different average urrents but for a onstant amplitude, whih is higher than the ritial voltage. For both samples, the same breakdown data is plotted first as a funtion of period T, then as a funtion of pulse width t p.itis learly seen, that the breakdown urves almost oinide if they are plotted against the pulse width, suh that the breakdown starts at the same ritial pulse width for different periods i.e., for different average urrents and pauses between the pulses. This indiates the fat that the pause between the pulses has no important relevane. There is no aumulative effet of dissipation as long as the pause between two subsequent pulses is not shorter than a few nanoseonds, indiating a omplete relaxation of the pulse-exited eletrons bak to the QH state on this time sale. In other words the QHE breakdown is not a funtion of the period or of the duty yle, but is a funtion of the pulse width. FIG. 3. a I-V urves V SD 0.57 V for up sweep and b the average longitudinal ondutivity xx of a Corbino devie (C2) at filling fator 4. Average longitudinal resistivity xx and average urrent measured as a funtion of pulse width and pulse amplitude for a Hall bar sample (H3) at filling fator 4 (V SD 0.73 V). C. Corbino devies at Ä2 We have repeated these pulsed-field experiments on the Hall bar and Corbino devies at filling fator 2. The breakdown behavior of the Hall bars were qualitatively similar to the one at 4, but the Corbino devies revealed interesting differenes. As seen in Figs. 5 a and 5 b the breakdown for the Corbino devie is very abrupt at filling fator 2, and the QHE region is very stable almost with

4 SAǦOL, NACHTWEI, VON KLITZING, HEIN, AND EBERL FIG. 4. The QHE breakdown urves plotted as a funtion of period on the left and of the pulse width on the right for a onstant amplitude with different duty yles. Both a for the Hall bar (H3) and b for the Corbino devie (C2) the breakdown urves oinide if they are plotted against the pulse width. This means that the pause between the pulses does not play a signifiant role. out any noise in the xx signal, unlike the Hall bar and Corbino devies at filling fator 4. This differene may be aused by the edge effets 15,16 in the QH devies. There is only limited knowledge about the ourrene of edge states in Corbino rings, 17 but these states an be different from those of Hall bars, depending on the ontat material. This is, beause the edge states in Hall bars are determined by the depletion zone near the ethed boundaries of the sample, whereas in Corbino rings they are defined by the work funtion of the ontat metal. Depending on the work funtion of the metal with respet to the 2DES, different boundary zones between 2DES and metal, ranging from depletion to aumulation zones, are possible. A finite value of either xx in the Hall bar or of xx in the Corbino ring near the breakdown orresponds to an exhange of harges between the edges or ontats via the 2D bulk. At higher filling fators lower magneti fields, this proess seems to be rather similar in Hall bars and Corbino rings, due to smaller Landau energy gaps and orrespondingly higher tunneling probabilities for inter-landau-level transitions. At smaller filling fators higher magneti fields this tunneling probability dereases, and the breakdown may be stronger depending on the loal properties of the sample edge. Loal flutuations of the edge potential 17,18 ould serve as the starting point for dissipative urrent filaments see, e.g., Refs. 19 and 20. The evolution of suh filaments is a highly nonlinear phenomenon, i.e., small variations of the boundary onditions lead to drasti hanges of the resulting urrent distribution. The orresponding differene of the breakdown behavior of Hall bars and Corbino devies is, however, not ompletely understood yet. FIG. 5. Corbino devie C1 at 2. a The QHE breakdown urves plotted as a funtion of the period on the left and of the pulse width on the right for a onstant amplitude with different duty yles. b Average longitudinal ondutivity xx measured as a funtion of pulse width for different pulse amplitudes. I-V urves with V SD 1.1 V for up sweep and V SD 0.7 V for down sweep. Another differene was observed in the duty yle dependene of the Corbino devies at 2. In Fig. 5 a the breakdown urves are plotted as a funtion of the period on the left and as a funtion of the pulse width on the right. The urves also almost oinide if they are plotted against the pulse width, but in a different manner. Some urves oinide exatly at ertain ritial pulse widths t p. For example, the urves with duty yles 25% and 30% oinide exatly at t p 2.1 ns, while the urves with 10% and 20% oinide at t p 2.55 ns. The urve with 15% duty yle on the other hand breaks down at t p 2.4 ns. These jumplike hanges of the ritial pulse widths are one of the nonlinear effets, whih we observed at Corbino devies at filling fator 2. Apart from the duty yle dependene, the amplitude dependene Fig. 5 b also shows a different behavior. On the positive voltage side of the I-V urves in Fig. 5 the ritial voltages are V SD 1.1 V for the up sweep and V SD 0.7 V for the down sweep. But in Fig. 5 b the QHE breaks down at an amplitude of A 0.9 V, at the average value of these ritial voltage values, as if there is no hysteresis. As the same behavior was onfirmed at another Corbino devie (C2), we an onlude that the applied pulses average out the hysteresis in the I-V harateristis. The observed hysteresis and the asymmetry with respet to the urrent polarity in the I-V urves are more pronouned for 2 in omparison to 4. We attribute this to the higher energy gap between the Landau levels, giving rise to a redued tunneling probability and a redued number of dissipative urrent filaments. The evolution of these few filaments gives rise to a stronger pronouned dependene of the urrent distribution on small variations of the boundary onditions, whih an also depend on the polarity of the eletri field between the urrent ontats. Thus, the topology of the

5 TIME SCALE OF THE EXCITATION OF ELECTRONS... FIG. 6. The magneti field dependene of the pulse-indued QHE breakdown at a Corbino devie (C1) at 2. The loser to the integer filling fator the bigger is the ritial pulse width and the more abrupt is the breakdown. The inset shows the xx around filling fator 2. filaments annot be assumed to be idential for both bias polarities. At lower magneti fields higher filling fators, more dense networks of urrent paths are developed, leading to smoother I-V urves and less hysteresis or asymmetry. The magneti field dependenes of the xx vs t p and xx vs t p urves are, however, ommon in both Hall bar and Corbino geometries. If the magneti field is varied from its filling fator 2 position to higher or to lower values, the QHE breakdown ours in shorter times and more smoothly. As seen in Fig. 6, the loser the magneti field is to the integer filling fator the bigger is the ritial pulse width and the more abrupt is the QHE breakdown. This resembles the typial urrent-voltage harateristis of the QHE breakdown, in whih the ritial urrent density reahes its maximum value and the breakdown is more abrupt at integer filling fators. 3 IV. DISCUSSION A. A simple drift model It is possible to explain these results by a simple drift model. In the previous studies of Kaya et al., 6,7 the authors onluded that the relaxation and exitation of the eletrons near the QHE breakdown are due to the inter-landau-level transitions, whih are triggered by Coulomb sattering at impurities. We an onsider the ase for the Hall bars and assume that these impurities are distributed homogeneously throughout the sample as shown in Fig. 7 a. At an integer filling fator the eletrons drift with a onstant veloity v D along the Hall bar. Within a pulse width of t p the drift length is D v D t p. Sine the voltage level is set to zero within the interval between the pulses, the drifting of eletrons takes plae only within a pulse. As long as the pulse amplitude A is kept below the ritial d voltage V SD (A V SD ) there is no breakdown at all, and the system is FIG. 7. a Piture of the drifting distane of eletrons D.If D reahes the average sattering distane in, the breakdown ours. b The orresponding breakdown urve. The expeted linearity between the ritial pulse width and the inverse amplitude. The slope should give the ritial drift length. d Sheme of Landau levels in the presene of potential flutuations. Left: low Hall field, right: high Hall field. w: width of the sample L, R : hemial potentials at the left and the right of the sample always in the QH state, whatever the pulse width is. Even if the pulse amplitude is inreased above the ritial voltage (A V SD ), there is no breakdown for short pulses gray arrows in Fig. 7 a and the gray part of the breakdown urve in Fig. 7 b when t p t p so D D up to a ritial pulse width t p, beause within their drifting distanes D the eletrons do not enounter enough sattering events, whih are neessary for the breakdown. If the pulse width is long enough (t p t p ), the drifting distane of eletrons an reah the average distane between the satterers in ( D D in ), and the eletrons enounter enough sattering events and tunnel between Landau levels, starting the breakdown proess. Further inrease of the pulse width blak arrows in Fig. 7 a and the blak part of the breakdown urve in Fig. 7 b when D D in inreases the resistivity ausing the system to reah the omplete breakdown, marked by a saturation of the resistivity xx at Hall bars as shown in Fig. 7 b or by a saturation of the ondutivity xx at the Corbino rings. We an formulate these as follows: within a pulse at integer filling fators, the drift veloity v D is v D E H /B, 3 where E H is the Hall field, and B is the magneti field. Within a ritial pulse width t p, the ritial drift length D will be D v D t p. Assuming E H A/w for a homogeneous urrent flow valid for xx onst 0, xy onst inside the 2DES, see, e.g., Ref. 8, with A being the amplitude, and w being the width of the sample, we an formulate the ritial pulse width as t p D wb A

6 SAǦOL, NACHTWEI, VON KLITZING, HEIN, AND EBERL B is the ylotron energy, E s is the energy of the spin splitting is smaller at lower magneti fields, giving rise to a higher tunneling rate 1/ t : 21 1 t exp y2 4 B 2 exp 1 4 B 2 E w 2 ea, with B being the magneti length. Thus, the breakdown ours in shorter drifting distanes. Similarly, higher amplitudes A bend the Landau levels more, and derease the spatial separation y between them, ausing a higher tunneling rate and redution of the role of impurity-indued flutuations, as shown in Fig. 7 d. Lower mobility samples have more sattering enters, so that the tunneling rate between Landau levels is higher and therefore the ritial drift lengths for these samples are shorter. As seen in Fig. 8 b, the mobility dependene is more pronouned for smaller amplitudes, when the Landau-levels are less bent. As the real drift veloities, whih depend on the urrent distribution and the resulting effetive width of the urrent path, w eff w, are unknown, the values of D mark the lower limit for the drift distanes. However, the parameter dependenes remain valid, as disussed above. 5 FIG. 8. a The plots of ritial pulse width vs inverse amplitude and b ritial drift length vs amplitude b measured at Hall bar devies H1, H2, and H3 with different Hall mobilities and at different integer filling fators. Therefore we should expet a linear relation between the ritial pulse width and the inverse amplitude. In other words, the slope of the funtion t p f (1/A) should give us the ritial drift length D as shown in Fig. 7. Figure 8 a shows the ritial pulse widths t p measured at Hall bars of different mobilities as a funtion of the inverse pulse amplitude 1/A for filling fators 2, 4, and 6. For higher filling fators and higher amplitudes, the expeted linearity is almost ahieved. Moreover, the lower the mobility, the loser are the results to this linearity. This may be attributed to a more homogeneous urrent flow at lower mobilities and at higher filling fators. At lower filling fators and at higher mobilities, the funtion t p f (1/A) beomes nonlinear due to an inhomogeneous urrent flow indued by impurityrelated potential flutuations. In Fig. 8 b the ritial drift lengths as determined from Eq. 4 are plotted as a funtion of amplitude, for different filling fators and for different mobilities note that although the urves t p f (1/A) at 4 almost oinide for the devies H2 and H3, the orresponding traes D f (A) differ due to the differene in the eletron density n s and the resulting differene in the drift veloity at the QH plateau. The ritial drift lengths are smaller for higher filling fators, beause the Landau-level energy gap E E s B. Corbino devies The ritial pulse widths in Corbino devies show, as disussed above, in general a similar behavior on the parameters as pulse amplitude and filling fator. However, a disussion of these parameter dependenes annot be based on similarly simplisti assumptions as in the ase of Hall bar devies. This is, beause even in the simplest ase of a homogeneous 2DES xx, xy onstant throughout the system, the potential distribution is nonlinear. The eletri field is highest near the inner eletrode, giving rise to the highest loal drift veloities. This should lead to an onset of dissipation near the inner eletrode. When the breakdown of the QHE ours, one or more urrent-arrying filaments may propagate from the inner to the outer eletrode, leading to a nonzero souredrain urrent. A similar behavior has been observed in thinfilm Corbino rings, prepared on n-doped GaAs. 19,20 There are some indiations of the formation of filaments at the breakdown of the QHE in our Corbino rings as listed below. 1 If the evolution of xx is plotted as a funtion of the pulse width, a distint differene in the behavior ompared to Hall bars is observed. At filling fator 2, the xx (t) urves of the Corbino rings do not exatly oinide for different duty yles t p /T, but group in ertain intervals of t p /T. Within one group of traes, t p is preisely idential, and the transition from QH to dissipative ondution is abrupt, in ontrast to Hall bars. Thus, eah group of traes may orrespond to a ertain number and onfiguration of dissipative urrent filaments between entral and outer eletrode, as observed diretly by spae-resolved photoluminesene in thinfilm Corbino rings. 19,20 The small values of xx 0 before the breakdown and the steepness of the transition at 2 indiate that the Corbino ring is a very effetive insulator in the QH state until a very sudden onset of dissipation. The origin of this behavior is not yet understood

7 TIME SCALE OF THE EXCITATION OF ELECTRONS... 2 Some of the xx t traes show more than one swithing point: not only for subsequent inrease and derease of the pulse width hysteresis, but also within single traes, subsequent swithing from the QH to the dissipative state and bak were observed when hanging pulse width t p. This may reflet the instability of the dissipative urrent filaments, due to the strong nonlinearity of the QH breakdown. However, at smaller magneti fields i.e., at 4, the abovementioned effets disappear, and the overall behavior of both Corbino ring and Hall bar devies tends to be very similar both xx and xx have small but nonzero values even for small exitation. V. SUMMARY To summarize, we have investigated the time sale of the exitation of eletrons in 2DES of GaAs/AlGaAs heterostrutures at the QHE breakdown. The measurements were performed by applying short eletri pulses to the QH devies with Hall bar and Corbino geometries at integer filling fators 2, 4, and 6, having different eletron mobilities ( H m 2 /V s). The breakdown starts only after a ertain ritial pulse width, if the amplitude of the pulses is higher than the ritial d voltage value. These ritial times are of the order of nanoseonds and are dependent on pulse amplitude, magneti field, and eletron mobility. From a simple model, whih assumes a homogeneous distribution of satterers and a homogeneous urrent flow throughout the sample, we obtain the lower limit for the ritial drift lengths varying from 1 to25 m. We interpret the ritial pulse widths as the average drifting times of the eletrons between inelasti sattering events ausing tunneling between the Landau levels, and the ritial lengths as the average drifting distanes related to the inelasti sattering lengths in of the eletrons. Charateristi dependenes of these ritial drift lengths on the amplitude, filling fator, and the mobility have been observed and an be attributed to impurity-assisted inter-landau-level tunneling. The time sale of exitation ns is in agreement with an estimation from a previous work by Kawaguhi et al. 4 The authors introdued the onept of a swithing time T s for the eletron system to transit from the QHE regime to the dissipative breakdown regime, and estimated that it should be of the order of 10 ns. No general differenes of the response to short eletri pulses have been found between Hall bars and Corbino devies. The only deviations were observed in the behavior of the Corbino rings at filling fator 2 with their almost perfet QHE onditions and abrupt breakdown profiles, and with the jumps and bunhing effets in the ritial pulse widths. We believe that these effets are due to the different kinds of edge states in the Corbino rings, in omparison to Hall bars. ACKNOWLEDGMENTS The authors would like to thank Dr. N. Kalugin for tehnial help in the experiments. This work was supported by the DFG Shwerpunkt-programm Quanten-Hall-Systeme projet No. Na235/10-1 and B.E.S. was supported by the German Aademi Exhange Servie DAAD. 1 K. von Klitzing, G. Dorda, and M. Pepper, Phys. Rev. Lett. 45, K. Yoshihiro, J. Kinoshita, K. Inagaki, C. Yamanouhi, J. Moriyama, and S. Kawaji, Surf. Si. 113, G. Ebert, K. von Klitzing, K. Ploog, and G. Weimann, J. Phys. C 16, Y. Kawaguhi, F. Hayashi, S. Komiyama, T. Osada, Y. Shiraki, and R. Itoh, Jpn. J. Appl. Phys., Part 1 34, S. Komiyama, Y. Kawaguhi, T. Osada, and Y. Shiraki, Phys. Rev. Lett. 77, I. I. Kaya, G. Nahtwei, K. von Klitzing, and K. Eberl, Phys. Rev. B 58, R L. I. Kaya, G. Nahtwei, K. von Klitzing, and K. Eberl, Europhys. Lett. 46, G. Nahtwei, Physia E Amsterdam 4, L. Eaves and F. W. Sheard, Semiond. Si. Tehnol. 1, G. Nahtwei, I. I. Kaya, B. E. Sagol, K. von Klitzing, and K. Eberl, Physia B 272, K. Güven, R. R. Gerhardts, I. I. Kaya, B. E. Saǧol, and G. Nahtwei, Phys. Rev. B 65, V. Tsemekhman, K. Tsemekhman, C. Wexler, J. H. Han, and D. J. Thouless, Phys. Rev. B 55, R B. E. Sagol, G. Nahtwei, I. I. Kaya, K. von Klitzing, and K. Eberl, in Proeedings of the 25th International Conferene on the Physis of Semiondutors, Osaka-2000, edited by N. Miura and T. Ando, Springer Proeedings in Physis Springer, New York, 2001, Vol. 87 Pt. II, p S. Komiyama, T. Takamasu, S. Hiyamizu, and S. Sasa, Solid State Commun. 54, M. Büttiker, Phys. Rev. B 38, R. J. Haug, Semiond. Si. Tehnol. 8, E. Ahlswede, J. Weis, K. von Klitzing, and K. Eberl, Physia E Amsterdam 12, R. P. Taylor, R. Newbury, A. S. Sahrajda, Y. Feng, P. T. Coleridge, M. Davies, and J. P. MCaffrey, Superlatties Mirostrut. 24, W. Eberle, J. Hirshinger, U. Margull, W. Prettl, V. Novak, and H. Kostial, Appl. Phys. Lett. 68, K. Aoki and S. Fukui, Physia B 272, T. Martin and S. Feng, Phys. Rev. Lett. 64,